At 10/19/12 10:45 AM, Innermike wrote: Probably nothing as interesting as you. My last flash game failed to sell after ages on FGL and hundreds of sponsor emails, so I deviated off from flash to trying my hand at iOS development and I've ended up landing several different contracts just making in house apps for small businesses like flower shops for stock managing etc. Most of them being nearly identical in functionality and equally unnecessary, I almost felt like I was scamming people.

okay, i'm pretty damn impressed.

I've always been adamantly against using ready made libraries when programming, I know how the old argument goes "Why reinvent the wheel?

having been on both sides of this whole thing (against and then for) I can honestly say that while true appreciation for libraries does come from having tried doing it yourself, the creator of that library only really wants people to use it.

I have tried to re-create many well-known libraries because I thought it was a fun learning experience. Now that i'm past that, i'll pick up a new library and a few tutorials on it without any regrets.

That moment in Java class when someone asks how to copy an image to a JPanel without having to readd everything else, and you want to scream "COPYPIXELS DAMMIT."

That moment in Java class when you ask how to get around the half second delay between pressing a key and holding it, but your prof doesn't notice it and thinks you're talking about the framerate of keyevents.

I type as Java begins updating itself

I wasn't too fond of using libraries, but TweenLite and NAPE are too good. For other stuff, I usually make it myself. I wasn't about to make my own isometric engine though...

At 10/19/12 11:14 AM, MSGhero wrote: That moment in Java class when someone asks how to copy an image to a JPanel without having to readd everything else, and you want to scream "COPYPIXELS DAMMIT."

"COPYPIXE-oh" xD

I type as Java begins updating itself

Every compile in Eclipse it just gets slower and slower and slower until it's eventually completely unusable and takes 5 seconds to type a single character. Then I have to close and re-open Eclipse.
You have the same issue?

It's the most annoying thing in the world, especially since it affects opening other windows as well. It all just gets slower without even attempting to use any of my CPU.

At 10/19/12 11:24 AM, egg82 wrote: Every compile in Eclipse it just gets slower and slower and slower until it's eventually completely unusable and takes 5 seconds to type a single character. Then I have to close and re-open Eclipse.
You have the same issue?

Nope, only the first compile takes a while, but that's normal. Everything after that takes as long as it should.

At 10/19/12 11:37 AM, MSGhero wrote: Nope, only the first compile takes a while, but that's normal. Everything after that takes as long as it should.

why does Eclipse hate me? Q.Q

I heard it's an issue with Apple computers, but... Well, I don't use Mac.

At 10/19/12 11:41 AM, egg82 wrote: why does Eclipse hate me? Q.Q

Eclipse got bloated over the years. But still its waaay faster on linux (at least on my computer). The autocomplete is as fast as FlashDevelops on Windows. Eclipse on Windows for me always needs a noticeable amount of time to load up the autocomplete list.
Still there are always tricks to speed it up I guess.

At 10/20/12 12:08 PM, Innermike wrote: Quick question:
I'm going to make a pixel artist request soon in the collaboration forum, but before I expose my current game engine to scrutiny I was wondering, what do you guys think about a flamethrower as a weapon? I was thinking of using it as a game mechanic (lighting torches, melting ice etc) but I have a feeling it's a little too sloppy for combat, do you guys prefer a more straightforward projectile type deal or would you be interested in something a little more loose?

Also, can we start an argument about how 'haXe' is pronounced please?

What are you looking for in combat? A flamethrower is a pretty neat idea as a game mechanic; maybe you should change combat to be able to fit the flamethrower more. Or give certain enemies fire-retardant suits or something.

Hex, but everyone says hacks.

At 10/20/12 12:13 PM, MSGhero wrote: Hex, but everyone says hacks.

I pronounce it "haycks"

At 10/20/12 12:08 PM, Innermike wrote: Also, can we start an argument about how 'haXe' is pronounced please?

So mike, before you took your break from the reg lounge you were saying how you want to make a game that is meaningful to you - something of satisfactory depth. I see you have some works in development. Have you made progress with that resolution?

its hakes whymes with cakes

thats how i pronounce it

Been away from flash for awhile. Have a game that is almost complete. I just need to add an ending cut scene and sort out some minor things. Just saying hi, and that I hope to become active if school work does not get in the way.

-Travis

I just made a game system that creates a map grid, randomly fills it with buildings, glues any walls that're too close together, beziers corners, and draws main roads between random points. Then I pasted a car into it and flash crashed and I didn't save :(

aaaaaaaaaa I was probably kiddin myself if I ever thought that could make a map as fun as doing one by hand.

At 10/21/12 10:48 AM, I-smel wrote: I just made a game system that creates a map grid, randomly fills it with buildings, glues any walls that're too close together, beziers corners, and draws main roads between random points. Then I pasted a car into it and flash crashed and I didn't save

Man, I hit Ctrl+S like every 3 seconds. It drives me nuts when I don't save for more than a minute.

so my ASWars bot is going pretty well. I finally figured out a way to get it to do exactly what I want.
That's all i'll say for now :P

At 10/21/12 11:26 AM, egg82 wrote: Man, I hit Ctrl+S like every 3 seconds. It drives me nuts when I don't save for more than a minute.

I always forget to save cuz I get used to not needing to save after compiling. After I have an "Oh shit, I haven't saved yet" moment, I subconsciously save. I also sync everything to Dropbox with SyncToy, which is really nice and easy to do.

so my ASWars bot is going pretty well. I finally figured out a way to get it to do exactly what I want.
That's all i'll say for now :P

I completely forgot about that lol. I have to study for a test tomorrow (and I'm lazy), so I might do some last minute AI.

At 10/21/12 01:00 PM, PSvils wrote: Wait...what's going on? :O

Get with the program, man. Tree's post a few down about AS Wars 2012

At 10/21/12 01:01 PM, MSGhero wrote: Get with the program, man. Tree's post a few down about AS Wars 2012

oh, yeah. I got it now. You're all going down >:D

my bot uses prediction, and knows exactly how each AI "thinks" and how the battle will play out.
It does not use classification, however, so predictions are as dynamic as the bots its predicting.

in other words, it's a learning AI.

At 10/21/12 01:28 PM, egg82 wrote: oh, yeah. I got it now. You're all going down >:D

my bot uses prediction, and knows exactly how each AI "thinks" and how the battle will play out.
It does not use classification, however, so predictions are as dynamic as the bots its predicting.

in other words, it's a learning AI.

*cough*tryhard*cough*

My bot may not be the best........ but its gonna look good dying!!!

At 10/21/12 02:29 PM, MSGhero wrote: *cough*tryhard*cough*

think you can beat it? ;)

so far my estimate is that it will be able to predict with a fair amount of accuracy up to 100 turns of the battle after being alive for 20 frames.

if you want to kill it, your best shot is to do so before then.

At 10/21/12 02:41 PM, egg82 wrote: think you can beat it? ;)

I think I'm just gonna make mine look cool like swish's.

At 10/21/12 02:41 PM, egg82 wrote:

At 10/21/12 02:29 PM, MSGhero wrote: *cough*tryhard*cough*

think you can beat it? ;)

so far my estimate is that it will be able to predict with a fair amount of accuracy up to 100 turns of the battle after being alive for 20 frames.

if you want to kill it, your best shot is to do so before then.

What is gonna be the name of your bot?? If it does what you say it does I wanna make a special case that goes and nukes your bot in the first 20 frames!!! lol.

At 10/21/12 02:43 PM, MSGhero wrote: I think I'm just gonna make mine look cool like swish's.

but... That wouldn't be a challenge D:

At 10/21/12 02:44 PM, swishcheese wrote: What is gonna be the name of your bot?? If it does what you say it does I wanna make a special case that goes and nukes your bot in the first 20 frames!!! lol.

so far i've named it H.U.N.T.R. -
Human-like
Unparalleled
and
Nefarious
Tracking
Rover

i'll probably come up with some better words for that anagram later.

coming directly at it in the first 20 frames isn't a good idea, though. It'll track that and beat the crap out of you. You'll have to go for something more unpredictable.

At 10/21/12 02:56 PM, egg82 wrote: coming directly at it in the first 20 frames isn't a good idea, though. It'll track that and beat the crap out of you. You'll have to go for something more unpredictable.

hmmm, maybe zig zagging. And what happens if you test your bots against each other?? Does one come on top based on firsts position or do the just battle forever??

At 10/21/12 03:06 PM, swishcheese wrote: hmmm, maybe zig zagging. And what happens if you test your bots against each other?? Does one come on top based on firsts position or do the just battle forever??

it tracks every bot's movement on every frame and uses pattern recognition to form predictions based on previous patterns. It does not have opinions on each pattern because it does not categorize patterns. It recognizes a pattern when it sees it and is able to predict that pattern's next move, but cannot attach a specific name to that pattern.

zig-zagging is a pattern. moving in a circle is a pattern. Moving toward or away from players and objects is a pattern. Moving in a circle, then a line, then a circle is a pattern.

the problem is, it takes two instances of that pattern to create a recognizable pattern. When it cannot predict a pattern, it will default to an AI "priority list" I created previously. That is its weak point.

the battle will last forever. This is only because I haven't gotten around to making decisions based on predictions yet.
basically it just stands there and analyzes everything.

At 10/21/12 01:28 PM, egg82 wrote: my bot uses prediction, and knows exactly how each AI "thinks" and how the battle will play out.

Good thing I read that, I was thinking about adding random elements to it from the start. Now we all know Math.random() is forbidden, but the great thing about pseudo random algorithms is that they are relatively random, but with the same seed each game it will play out exactly the same all the time.

I'd love to see my bot (which I have yet to make after the majority of my exams) screw up yours by making it believe mine uses strategies which aren't there :P.

At 10/21/12 03:18 PM, Sandremss128 wrote: I'd love to see my bot (which I have yet to make after the majority of my exams) screw up yours by making it believe mine uses strategies which aren't there :P.

good job! Pseudo random numbers would completely screw with my bot's pattern recognition until it was able to see the larger pattern that the pseudo-random generates. That'll buy you some more time.

At 10/21/12 03:39 PM, PSvils wrote: Which I doubt is likely...

it'll eventually see a pattern, because there is a pattern there. It may take a long-ass time, though :P

the pattern recognition I created works like this:
it groups similar numbers together and looks for patterns within those groups. If it finds a pattern, it makes a note of it and keeps going. Eventually it winds up with a vector of usable number patterns. The only issue after that is finding out which order the patterns persist, and making a prediction based on the current position of the bot and that set of patterns.

how does it group numbers?
it rounds the numbers in the main vector and looks between some simple number buffers.

how does it find a pattern?
simple math. Looking for addition and subtraction between direction a and direction b and averaging it out.

for example: say your bot moves in a circle (example bot)
it calculates the direction between point a and point b on each frame (the last position of the bot and the current position of the bot) - each directional movement between the two points increments by 2 degrees (just guessing), which is well between the buffers (the buffers grow if there's too many pattern groups) - and I end up with: 2, 2, 2, 2, 2, etc.
once it sees this pattern, it predicts the next point will be add 2 degrees (and 5 pixels) from where it is now.

now let's say example bot's circle algorithm grows:
I end up with: 2, 4, 6, 8, 10
4-2 = 2, 6-4 = 2, 8-6 = 2, etc.
and once again I can predict where it's going to land next.

of course, this works with multiple patterns as well:
a circle is 2, 2, 2, 2, etc.
a line is 0, 0, 0, 0, etc.
a circle is 2, 2, 2, 2, etc.
2, 0, 2 - means I need to get the average of how long it travels in a circle (let's say 10 frames) and get the next number in the pattern. If we happen to be on frame 1 of the new pattern, it's probably a 2. If we're on frame 10, the next one is probably a 0.

So, egg82: What if our bots are positioned facing each other and your bot starts firing at mine, my bot doesn't move until the very last second to dodge the bullet, does this mean that all the current bullets on the screen will already miss (so far the 'pattern' of my bot would have been to not move)? If so you're at a considerable disadvantage already.
I'm not making a bot that does the same pattern every time, it will react on your moves and by the time that you've cracked the pattern (which I doubt to what extend you're able to do this, it might actually be more complex than making a strong bot) your bot is at such a disadvantage mine will be able to take your fire because of the advantage in resources it has accumulated.

I'm just interested to know what your strategies are considering this.

Also the pattern you've build during the 4 or 3 bot FFA (free for all) might become obsolete as I have very different ideas about each situation (4 FFA, 3FFA, and 1v1).

This bot... I need some aspirin -.-'

At 10/21/12 04:27 PM, Sandremss128 wrote: So, egg82: What if our bots are positioned facing each other and your bot starts firing at mine, my bot doesn't move until the very last second to dodge the bullet, does this mean that all the current bullets on the screen will already miss (so far the 'pattern' of my bot would have been to not move)? If so you're at a considerable disadvantage already.

I thought about something like that before. My initial strategy was to categorize the different behaviors of bots, and i'll be using some of that here. I know I said there's no categorization with my bot, so i'll amend that to: there's very little categorization with my bot.

A bullet dodge is a bullet dodge is a bullet dodge to my bot.

I'm not making a bot that does the same pattern every time, it will react on your moves and by the time that you've cracked the pattern (which I doubt to what extend you're able to do this, it might actually be more complex than making a strong bot) your bot is at such a disadvantage mine will be able to take your fire because of the advantage in resources it has accumulated.

again, this is where the basic categorization comes in. Here are the categories:
away from player
toward payer
away from mine
toward mine
away from bullet
toward bullet
time

minus looking at opposites, each one is a separate category and each has its own point value. Decisions about attacking are based partially on this.

I'm just interested to know what your strategies are considering this.

mhm, I want my bot to get beaten :3
(but of course, i'm not going to make it easy)

Also the pattern you've build during the 4 or 3 bot FFA (free for all) might become obsolete as I have very different ideas about each situation (4 FFA, 3FFA, and 1v1).

the bot was designed with different situations in mind. It handles any type of situation well, as long as some basic rules stay the same:
1. That there's another bot of any kind on the screen.
2. That there's bullets
3. That there's mines
4. That there's no other additions to the game

remember: it analyzes and predicts on-the-fly, so changing strategies isn't an issue.

also, be aware that tracking and predicting movement isn't going to be its only trick. It tracks and analyzes literally everything that happens.

I got a headache trying to do all this prediction stuff, so i'm going to do something easier for the rest of the day. I feel i've made enough progress to call it a day on that project, anyway.

maybe i'll get back to work on loading 3D objects from a server, or even go for my dynamic mesh idea.